Probabilistic orbits and dynamical masses of emission-line binaries

The observed orbits of emission-line stars may be affected by systematics owing to their broad emission lines being formed in complex and extended environments. This is problematic when orbital parameter probability distributions are estimated assuming radial-velocity data solely comprised Keplerian motion plus Gaussian white noise, leading overconfident inaccurate solutions, with implications for the inferred dynamical masses hence evolutionary models. We present a framework which these systems can meaningfully analysed. synthesise benchmark datasets, each different challenging noise formulation, testing performance algorithms. make datasets freely available aim making model validation an easy standardised practice this field. Next, we develop application processes binaries, named marginalised $\mathcal{GP}$. algorithm, along current algorithms, on synthetic find our $\mathcal{GP}$ algorithm performs significantly better than standard algorithms contaminated systematics. Finally, apply four prototypical binaries: Eta Carinae, GG WR 140, 133. several case studies; consequently, predicted distributions, masses, modified from those previously determined.